Assuming that your cues are compile-time constants you will have much more room to work with if you keep the cues in FLASH (PROGMEM) than if you use RAM. The ATmega328 has 32k of FLASH but only 2k of SRAM.
Figure out what units of measure you want to use and the ranges of values. Then you can minimize the data sizes. Remember that if you usuall need less than 255 steps you can use one byte for the number of steps and chain two motions together to get one longer motion.